Targeting Cell Membrane HDM2: A Novel Therapeutic Approach for Acute Myeloid Leukemia

Abstract

The E3 ligase human double minute 2 (HDM2) regulates the activity of the tumor suppressor protein p53. A p53-independent HDM2 expression has been reported on the membrane of cancer cells but not on that of normal cells. Herein, we first showed that membrane HDM2 (mHDM2) is exclusively expressed on human and mouse AML blasts, including leukemia stem cell (LSC)-enriched subpopulations, but not on normal hematopoietic stem cells (HSCs). Higher mHDM2 levels in AML blasts were associated with leukemia initiating capacity, quiescence and chemoresistance. We also showed that a synthetic peptide PNC-27 binds to mHDM2 and enhances interaction of mHDM2 and E-cadherin on the cell membrane, in turn E-cadherin ubiquitination and degradation, leading to membrane damage and cell death of AML blasts by necrobiosis. PNC-27 treatment in vivo resulted in a significant killing of both AML “bulk” blasts and LSCs, as demonstrated respectively in primary and secondary transplant experiments using both human and murine AML models. Notably, PNC-27 spares normal HSC activity as demonstrated in primary and secondary BM transplant experiments of wild-type mice. We concluded that mHDM2 represents a novel and unique therapeutic target and targeting mHDM2 using PNC-27 selectively kills AML cells including LSCs, with minimal off-target hematopoietic toxicity.

Introduction

Acute myeloid leukemia (AML) is a hematopoietic malignancy characterized by accumulation of primitive and partially differentiated clonal cells in bone marrow (BM), peripheral blood (PB), and extramedullary organs [e.g., spleen (SP)] resulting in disruption of normal hematopoiesis. High rates of relapse or refractory disease are likely due to persistence of so-called treatment invulnerable leukemia stem cells (LSCs).^1–3

Human double minute 2 (HDM2) is an E3 ubiquitin-protein ligase that, under physiologic conditions, induces ubiquitination and proteasomal degradation of p53, a transcription factor and tumor suppressor gene that regulates DNA-damage-induced apoptosis.⁴ Aberrant expression of HDM2, however, seemingly plays a p53-independent role in tumorigenesis.^5–11 While under physiological conditions, HDM2 is localized to the cytoplasm where it negatively regulates p53 activity, recently it has also been found to be expressed on the cell membrane of solid tumor cells (e.g., colon, breast, and pancreas)^12–16 and a myeloid leukemia cell line (K562).¹⁷ This observation suggested that a potentially exclusive membrane compartmentalization of HDM2 in malignant cells may occur and perhaps mediate a previously unrecognized oncogenic mechanism that offers a window of opportunity for specific therapeutic targeting of cancer over normal cells.

PNC-27 is a novel synthetic compound designed to target membrane (m) HDM2 (mHDM2). This synthetic 32-amino acid peptide contains two domains: an HDM-2-binding domain corresponding to residues 12–26 of p53 located on the amino terminus, and a trans-membrane-penetrating domain located on the carboxyl terminus.¹⁸ The cytotoxicity of PNC-27 appears to be p53-independent and uniquely related to its mHDM2 binding ability.^13,18 Once bound to mHDM2, PNC-27 adopts a strongly amphipathic helix-loop-helix (HLH) structure resulting in trans-membrane pore formation (i.e., poration)^15,19–21 and cell death via necrobiosis.^12–18,22 To date, however, neither mHDM2 expression nor PNC-27 anti-leukemia activity has been fully evaluated in AML.

Herein, we first show that mHDM2 is preferentially expressed on the surface membrane of AML cells, including LSC-enriched subpopulations, but not on that of normal hematopoietic stem cells (HSCs) or acute lymphoblastic leukemia (ALL) blasts, and associated with quiescence, leukemia initiating capacity, and chemoresistance. By targeting mHDM2, PNC-27 has a specific anti-AML activity in vitro and in vivo in both human and murine AML models. From a mechanistic standpoint, membrane poration in AML cells exposed to this compound is partially mediated by PNC-27-induced and mHDM2-dependent E-cadherin ubiquitination and degradation, which ultimately results in cell membrane damage and death.

Materials and Methods

Samples

Normal, AML and ALL samples were obtained from donors and patients at City of Hope National Medical Center (COHNMC) through COH IRB#18067 (see patients’ characteristics in Table S1). Sample acquisition and analysis was approved by the Institutional Review Boards at the COHNMC and met all requirements of the Declaration of Helsinki. Mononuclear cells (MNCs) were isolated using Ficoll separation and CD34⁺ cells were selected through magnetic beads (Miltenyi Biotech, Cologne, Germany).

Peptides

PNC-27 and negative controls PNC-29 and PNC-26 peptides were provided by Oncolyze, Inc (New York, NY, USA).^{12–14,16–18} 5-TAMRA fluorophore-labeled PNC-27 peptide was synthesized at Biopeptide Corp (La Jolla, CA, USA). ¹⁸

Cell culture and in vitro assays

Human CD34⁺ cells were cultured in Stemspan serum-free medium II (SFEM II, StemCell Technologies), supplemented with granulocyte-macrophage colony-stimulating factor (GM-CSF) 200pg/mL, leukemia inhibitory factor (LIF) 50pg/mL, granulocyte colony-stimulating factor (G-CSF) 1ng/mL, stem cell factor (SCF) 200pg/mL, macrophage-inflammatory protein-1α (MIP-1α) 200pg/mL, and interleukin-6 (IL-6) 1ng/mL.^23–25 Mouse BM Lineage⁻Sca-1⁺c-Kit⁺ (LSK) cells were cultured in SFEM II supplemented with 10ng/ml SCF and 10ng/ml TPO.²⁵ All cytokines were obtained from Pepro Tech US (Rocky Hill, NJ, USA).

Flow cytometric analysis

Cells were stained with rabbit anti-HDM2 antibody (sc-813, Santa Cruz Biotechnology, Dallas, TX, USA) or normal-rabbit-IgG (sc-2027, Santa Cruz Biotechnology) antibody as control, and then with secondary Alexa Fluor 488-conjugated donkey anti-rabbit IgG antibody (Life technologies, Eugene, OR, USA), followed by assessment for cellular membrane HDM2 expression on cell surface by LSRII flow cytometer (BD Biosciences, San Diego, CA, USA). Mouse BM Lin⁻ cells were selected using mouse Lineage depletion microbeads (Miltenyi Biotech). Human CD34⁺HDM2^high, CD34⁺HDM2^low cells and mouse LSK cells were obtained by fluorescence-activated cell sorting (FACS) on ARIAIII or SORP (BD Biosciences).

Animal studies

The Mll^PTD/WT/Flt3^ITD/ITD mouse (CD45.2, B6 background) was selected as a murine AML model for our studies.²⁶ Mll^PTD/WT/Flt3^ITD/ITD CD45.2 AML cells were transplanted in CD45.1 B6 mice (Charles River, Wilmington, MA, USA) to obtain a synchronous cohort of leukemic mice and allow tracking of donor CD45.2 AML cells after transplantation. Human AML blasts were transplanted into NOD.Cg-PrkdcscidII2rgtm1WjlTg (CMV-IL3, CSF2, KITLG)1Eav/MloySzJ mice (NSG-SGM3, The Jackson Laboratory, Bar Harbor, ME, USA) to obtain patient sample-derived xenografts (PDXs). Mouse care and experimental procedures were performed in accordance with protocols approved by the COHNMC Institutional Animal Care and Use Committee.

Engraftment of human cells in immunodeficient mice

AML HDM2^high and HDM2^low blasts selected by FACS and transplanted via tail vein injection into 6–8-week-old NSG-SGM3 mice (300cGy) which were monitored for PB AML engraftment (hCD45⁺) every 4 weeks (wks) and survival.

In vivo treatment of primary human AML xenografts

AML blasts from AML patients were transplanted (2×10⁶ cells/mouse) into 6–8-week-old NSG-SGM3 mice (300cGy) via tail vein injection. After engraftment was confirmed, the mice were treated for 3wks with PNC-27, PNC-29 (40mg/kg, ip, daily) or vehicle, and monitored for survival. Another cohort of recipients was treated 2wks with 40mg/kg or 100mg/kg PNC-27, PNC-29 (ip, daily) or vehicle. Human AML engrafted cells (hCD45⁺ and hCD45⁺CD34⁺) were analyzed. BM cells from treated mice were harvested, pooled and transplanted into another cohort of NSG-SGM3 recipients (2×10⁶ cells/mouse, 300cGy), which were monitored for AML cell engraftment (hCD45⁺) and survival.

Plasma membrane isolation

Total plasma membrane proteins of AML and normal cells were extracted using a plasma membrane protein extraction kit, according to manufacturer’s protocol (ab65400, Abcam, Cambridge, MA, USA).

Immunoprecipitation (IP)

Antibodies used for IP were conjugated with protein A/G agarose beads using an antibody crosslinking kit (Santa Cruz Biotechnology).²⁵

Statistics

All in vitro experiments were performed in triplicate or more. For in vivo treatment experiments, animal numbers were chosen based on experimental group size, mice availability and variation, and treatment frequency. All statistical analyses were conducted using prism software version 7.0 (GraphPad Software, La Jolla, CA, USA). Results are shown as mean ± s.e.m. Statistical significance of differences was measured between groups using student’s t-test. Kaplan-Meier survival curves were used to display the overall survival results, and log-rank test was used to assess significant differences between survival curves. A two-sided P-value ˂0.05 was considered significant (Legend: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001).

Additional detailed methods are provided in the supplemental materials and methods.

Results

HDM2 is selectively expressed on the cell surface of AML blasts

Expression of mHDM2 was examined in AML and ALL cell lines by flow cytometry. All but one (i.e., KG1A) of the AML cell lines and none of ALL cell lines exhibited mHDM2 expression (Figure S1A). The median immunofluorescence (IF) intensity (MFI, relative to IgG controls) of mHDM2 was 2 to 4-fold higher in AML cells (except KG1A) compared to that of ALL cells (Figure S1B–C). mHDM2 expression was also significantly higher in MNCs from AML patients (n=8) compared to MNCs from ALL patients (n=3) and healthy donors (n=8) (Figure 1A–D) as measured by flow cytometry, immunoblot (IB) assays or IF staining.

Figure 1. HDM2 is selectively expressed on the cell surface of AML blasts and associated with leukemogenic capacity, quiescence and chemoresistance.

A-B. Representative HDM2 expression (A) and relative median fluorescence intensity (MFI) (HDM2/IgG) (B), in cells from AML (n=8) and ALL (n=3) patients and MNCs from healthy donors (n=8), assessed by flow cytometry analysis (red). Normal rabbit IgG stained as controls (black). C. Membrane HDM2 expression level was assessed by IB assays. Membrane fraction of normal or AML cells was immunoblotted with indicated antibodies. Anti-Na, K-ATPase and anti-GAPDH antibodies were used as positive and negative markers for determination of purity of membrane extract respectively. D. HDM2 (green) and cellular membrane (red), as assessed by staining with anti-HDM2 and WGA antibody respectively in MNCs from AML patients or healthy donors, yellow signal indicates co-localization. Representative figures are shown. Scale bar, 5μm. E-F. Representative HDM2 expression assessed by flow cytometry (E) and relative MFI (HDM2/IgG) (F) in immunophenotypically defined subpopulations (CD34⁻, CD34⁺, CD34⁺CD38⁺, CD34⁺CD38⁻ cells) from consecutive AML patients (n=10) and healthy individuals (n=8). G. FLT3-ITD⁺ (AML1) and H. FLT3-ITD^- (AML 10) AML blasts were selected at the high (mHDM2^high) and low (mHDM2^low) 20^th percentiles of HDM2 expression by flow cytometry (left panels). HDM2^high and HDM2^low AML blasts were respectively transplanted into NSG-SGM3 mice recipients (AML1: n=6 each; AML 10: n=3 each) and followed for engraftment in PB (middle panels) and survival (right panels). Results shown represent mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001; NS, not significant; by two-tailed, paired student’s t-test. The log-rank test was used to assess significant differences between survival curves.

mHDM2 expression was found both on CD34⁺ (CD34⁺CD38⁻ and CD34⁺CD38⁺) and CD34⁻ AML blasts (n=10) but not on the relative counterparts from healthy donors (n=8) (Figure 1E–F). Among AML cell subpopulations, mHDM2 expression was significantly higher in CD34⁺ (P<0.05) and CD34⁺CD38⁻ (P<0.05) cells compared to CD34⁻ cells, suggesting increased expression of mHDM2 in more primitive leukemia cells (Figure 1F, Figure S1D).

High mHDM2 expression associates with enhanced leukemogenic capacity, increased quiescence and chemoresistance in AML blasts

Next, we selected primary AML blasts in the high (mHDM2^high) and low (mHDM2^low) 20^th percentiles by flow cytometry from three representative FLT3-ITD⁺ (AML1), FLT3-ITD⁺ (AML3), FLT3-ITD⁻ (AML10) AML patients (Figure 1G–H, left panels). The cell cycle status of mHDM2^high vs mHDM2^low AML blasts was evaluated by Ki-67 and DAPI labeling. We observed a significantly increased proportion of cells in G_o phase in mHDM2^high compared with mHDM2^low AML blasts, indicating an increased proportion of quiescent cells in the mHDM2^high AML subpopulation (Figure S1E). In addition, mHDM2^high AML blasts were more chemoresistant compared with mHDM2^low AML blasts, as corroborated by reduced apoptosis (Figure S1F) upon exposure to cytarabine (Ara-C) or daunorubicin (DNR).

To determine in vivo leukemia-initiating capacity, mHDM2^high and mHDM2^low AML blasts from two AML patients, respectively FLT3-ITD⁺ (AML1) and FLT-3-ITD⁻(AML10), were transplanted into NSG-SGM3 mice. Recipient mice transplanted with mHDM2^high AML blasts had a significantly higher PB AML cell engraftment (P<0.001, Figure 1G–H, middle panels) and a significantly shorter survival (median survival in AML1: 114 vs 144.5 days, P<0.01; in AML10: 40 vs 52 days, P<0.05, Figure 1G–H, right panels) than recipients transplanted with mHDM2^low AML blasts.

Anti-leukemia activity of PNC-27 in vitro by targeting mHDM2

Since mHDM2 was selectively expressed on AML blasts, we then hypothesized that it may represent a unique therapeutic target. PNC-27 is a synthetic 32-amino acid peptide containing a p53 protein sequence that enables binding to mHDM2. PNC-27 killing activity has been previously reported in solid tumors^{12–16,18,22}, but little is known regarding its activity in AML. Thus, we first incubated AML and normal CD34⁺ blasts with 5-TAMRA fluorophore-labeled PNC-27 peptide for 4 hours (hrs) and then stained them with anti-HDM2 antibody¹⁸. We confirmed co-localization of mHDM2 and PNC-27 to the membrane of AML CD34⁺ blasts with no or minimal co-localization to that of normal CD34⁺ cells (Figure 2A). To determine the antileukemia activity of PNC-27, we then exposed primary mHDM2-positive CD34⁺ blasts (MFI>1.5) from AML patients (n=12) to PNC-27 and to control peptides PNC-26 or PNC-29 (concentration range 0–37.5μM) for 24, 48 and 72hrs. We observed a significant dose-dependent decrease in cell viability in primary CD34⁺ blasts from AML patients (n=12) treated with PNC-27 but not in those treated with PNC-26 or PNC-29 (Figure 2B, Figure S2A–B). The IC50 for primary mHDM2-positive AML blasts ranged between 11.64 to 31.64 μM at 48hrs. Only minimal cell killing was observed in PNC-27-treated mHDM2-negative CD34⁺ cells from healthy donors (n=9) (Figure 2B, Figure S2A–B). Similar differences were observed in mHDM2-positive (i.e., MV4-11, MOLM13, THP-1 and NOMO-1) vs mHDM2-negative (KG1A) AML cell lines, with a PNC-27 IC50 ranging between 5.83 to 18.54 μM at 48hrs (Figure S2C–K). A significant decrease in colony forming capacity (CFC) was also observed in primary AML CD34⁺ blasts, but not in normal CD34⁺ cells exposed to PNC-27 (25μM, 72hrs) (Figure 2C–D) or in primary AML CD34⁺ blasts exposed to PNC-26 or PNC-29 (25μM, 72hrs) (Figure 2C), suggesting a specific targeting of AML blasts. A significant positive correlation between mHDM2 expression levels and PNC-27 anti-leukemia activity in primary AML CD34⁺ blasts (Figure 2E) was also observed.

Figure 2. In vitro activity of PNC-27.

A. Co-localization (yellow) of HDM2 and 5-TAMRA fluorophore-labeled PNC-27 (red) in AML CD34⁺ cells, as assessed by staining with anti-HDM2 (green) antibodies in AML CD34⁺ and normal CD34⁺ cells with and without PNC-27 (25μM, 4hrs). Representative figures are shown. Scale bar, 5μm. B: Viability of mHDM2-positive primary AML CD34⁺ cells (n=12) and mHDM2-negative normal CD34⁺ cells (n=9) at 48hrs after treatment with PNC-27, PNC-29 and PNC-26 (0–37.5μM) as assessed by viability assays. C-D: Colony forming capacity after exposure to PNC-27, PNC-26, PNC-29 (25μM, 72hrs) in primary AML CD34⁺ cells (n=9) (C) and colony forming capacity after exposure to PNC-27 (25μM, 72hrs) in normal CD34⁺ cells (n=6) (D). E: Correlation of IC-50s at 48hrs with MFI of mHDM-2 expression in primary AML cells (n=16). F. Co-localization (yellow) of HDM2 and 5-TAMRA fluorophore-labeled PNC-27 (red), as assessed by staining with anti-HDM2 (green) antibodies in LSK cells from Mll^PTD/WT/Flt3^ITD/ITD AML mouse and wt mouse with and without PNC-27 (25μM, 4hrs). Representative figures are shown. Scale bar, 5μm. G. Viability of AML LSK cells (n=5) and wt LSK cells (n=3) at 48hrs after treatment with PNC-27, PNC-29 and PNC-26 (0–37.5μM) as assessed by viability assays. Results shown represent mean ± s.e.m. ***P<0.001, ****P<0.0001; NS, not significant; by two-tailed, paired student’s t-test.

We then evaluated these findings in the Mll^PTD/WT/Flt3^ITD/ITD AML mouse model, which recapitulates human FLT3-ITD AML.²⁶ Similar to human AML, we observed expression of mHDM2 on the membrane of Mll^PTD/WT/Flt3^ITD/ITD AML LSK but not on that of wild type (wt) LSK cells. Furthermore, PNC-27 and mHDM-2 co-localized to the cell membrane of AML LSK cells but not to that of wt LSK cells (Figure 2F). Consistent with these results, we observed a significantly reduced cell viability in PNC-27-treated AML LSK cells and not in PNC-27-treated wt LSKs or PNC-26 or PNC-29- treated AML LSK cells (Figure 2G).

PNC-27 antileukemia activity in vivo

We then tested the antileukemia activity of PNC-27 in vivo. CD45.2 BM cells from Mll^PTD/WT/Flt3^ITD/ITD AML mice were transplanted into CD45.1 syngeneic wt recipients to generate a synchronous cohort of leukemic mice. After confirming engraftment and AML development (WBC>20×10⁹/L), the leukemic mice were treated with PNC-27, PNC-29 or vehicle (see Methods). One cohort of mice was followed for survival, and another sacrificed at the end of treatment for evaluation and harvesting of BM, PB and SP (Figure 3A). The selected PNC-27 dose and treatment schedule were similar to those previously reported for in vivo experiments in solid tumors.¹⁶ We observed a prolonged survival in PNC-27 treated mice (n=9), compared to PNC-29- (n=10) or vehicle-treated mice (n=10) (median survival: 54 vs 41 vs 42 days, P_{PNC-27 vs PNC-29}=0.021, P_{PNC-27 vs vehicle}=0.028, Figure 3B). We also observed significantly reduced CD45.2⁺ AML cell counts in BM, PB and SP (Figure 3C–E) as well as decreased CD45.2⁺ AML LSK cell counts in BM (Figure 3F) in PNC-27-treated mice (n=6), compared to PNC-29-treated (n=5) or vehicle-treated controls (n=6). Since we were interested in assessing the activity of PNC-27 on LSC-enriched fractions of AML blasts, BM cells from PNC-27- or PNC-29- or vehicle-treated AML mice were transplanted into CD45.1 recipient mice. Secondary recipients of BM cells from PNC-27-treated donors (n=6) showed a significantly reduced fraction of circulating CD45.2⁺ AML cells (Figure 3G) and a significantly prolonged survival (median survival: 43 vs 32.5 vs 32 days, P_{PNC-27 vs PNC-29}=0.0021, P_{PNC-27 vs vehicle}=0.0009, Figure 3H) compared to recipients of BM cells from PNC-29- (n=8) or vehicle-treated donors (n=6).

Figure 3. In vivo activity of PNC-27 in Mll^PTD/WT/Flt3^ITD/ITDAML mouse model and human AML patient sample-derived xenografts (PDXs).

A-H: Effect of PNC-27 and PNC-29 in Mll^PTD/WT/Flt3^ITD/ITDAML mouse model. A. Experimental design of the mouse AML model. CD45.2 BM cells from leukemic mouse were transplanted into CD45.1 congenic mice to generate a synchronous cohort of leukemic mice. After confirming engraftment and AML development, the mice were treated with vehicle, PNC-29 or PNC-27 (40mg/kg, ip, daily) for 3wks and followed for survival. Another cohort of leukemic mice were treated with vehicle, PNC-29 or PNC-27 (40mg/kg, ip, daily) for 2wks and followed for leukemia stem cells burden analysis after BM cells were harvested and transplanted into secondary CD45.1 recipient mice. B. Survival in vehicle- (n=10), PNC-29- (n=10) or PNC-27- (n=9) treated primary recipients. C-F. Number of CD45.2 cells in BM (C), PB (D) and SP (E) and CD45.2 AML BM LSK cells (F) from vehicle- (n=6), PNC-29- (n=5) or PNC-27- (n=6) treated mice. G-H. CD45.2 AML cell engraftment in PB at 4wks (G) and survival (H) of secondary recipient mice receiving BM from vehicle- (n=6), PNC-29- (n=8) or PNC-27- (n=6) treated donors. I-O. Effect of PNC-27 and PNC-29 in AML PDXs. I. Experimental design of the PDX study. NSG-SGM3 mice were transplanted with BM blasts from an AML patient (AML1). After AML engraftment was confirmed, the mice were treated with vehicle, PNC-29 or PNC-27 (40mg/kg, ip, daily) for 3wks and followed for survival. Another cohort of leukemic mice were treated with vehicle, or PNC-27 (40mg/kg, ip, daily) for 2wks and followed for leukemia stem cells burden analysis after BM cells were harvested and transplanted into secondary NSG-SGM3 recipient mice. J-M. AML cell engraftment in PB (J) and survival (K) in vehicle- (n=7), PNC-29- (n=8) or PNC-27- (n=8) treated primary recipients. hCD45⁺CD34⁺ engraftment in BM (L) and hCD45⁺ engraftment in SP (M) in PNC-27 (n=6) or vehicle (n=5) treated recipients. N-O: hCD45⁺ engraftment in PB measured every 4wks after transplantation (N) and survival (O) of recipient mice receiving BM cells from PNC-27-treated (n=10) or vehicle-treated mice (n=7). Results shown represent mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; NS, not significant; by two-tailed, paired student’s t-test. The log-rank test was used to assess significant differences between survival curves.

Similar results were observed in human AML PDXs. We transplanted NSG-SGM3 mice with human AML blasts (AML1) and treated them with PNC-27, PNC-29 or vehicle, after confirming engraftment and disease. One group was then followed for survival and another was sacrificed at the end of treatment, BM, PB and SP were evaluated for leukemia burden (Figure 3I). We observed reduced AML engraftment (Figure 3J) and prolonged survival in PNC-27-treated primary recipient mice (n=8), compared with PNC-29- (n=8) and vehicle-treated (n=7) controls (median survival: 136.5 vs 115 vs 111 days; P_{PNC-27 VS PNC-29}=0.0034, P_{PNC-27 VS vehicle}=0.0014; Figure 3K). Significantly reduced engraftment in BM (Figure 3L) and SP (Figure 3M) were also observed in the PNC-27-treated mice (n=6) compared with vehicle-treated mice (n=5). BM cells harvested from PNC-27 or vehicle-treated mice were then transplanted into secondary NSG-SGM3 recipients to assess LSC burden. Reduced hCD45⁺ cell engraftment (Figure 3N) and significantly prolonged survival were observed in recipient mice receiving BM cells from PNC-27-treated donors (n=10) compared with those receiving BM cells from vehicle-treated donors (n=7) (median survival: 156.5 vs 93 days; P<0.0001; Figure 3O).

PNC-27 has minimal off-target toxicity on normal hematopoiesis

Since mHDM2 is expressed at very low levels or totally absent in normal HSCs, we expected little activity of PNC-27 on normal hematopoiesis. To verify this hypothesis, we treated normal healthy B6 mice with 100mg/kg PNC-27 (ip, daily, a higher dose than that used to demonstrate PNC-27 antileukemia activity) or vehicle for 2wks, then we harvested PB, BM and SP for analysis (Figure S3A). We observed no significant differences between the two groups (n=7 in each) across a variety of hematologic parameters including WBC, HB and PLT counts in PB (Figure S3B–D), BM and SP total MNCs (Figure S3E–F) and BM LSK cell percentage and counts (Figure S3G–H). To assess the effect of PNC-27 on hematopoietic reconstitution ability, BM long-term HSCs (LT-HSCs, Flt3⁻CD150⁺CD48⁻ LSK) cells from PNC-27- or vehicle-treated healthy mice were transplanted into CD45.1 recipient mice. No significant differences on WBC counts (Figure S3I) or donor CD45.2⁺ engraftment (Figure S3J) were observed between recipient mice receiving BM cells from PNC-27-treated donors (n=9) and those receiving BM from vehicle-treated donors (n=9) at 16wks follow-up. In contrast, we observed a significantly reduced hCD45⁺ cell engraftment in BM, PB and SP (Figure S4A–D), and significantly reduced hCD45⁺CD34⁺ cell engraftment in BM and SP (Figure S4E–F) in the NSG-SGM3 mice transplanted with human AML blasts (AML2) and treated with at the same higher dose of PNC-27 (100mg/kg, ip, daily, 2wks). BM cells from the PNC-27- or vehicle-treated mice were then transplanted into recipient NSG-SGM3 mice. At 16wks, we harvested PB, BM, and SP cells for analysis. Significantly reduced hCD45⁺ (Figure S4G) and hCD45⁺CD34⁺ cell engraftment (Figure S4H) in BM was observed in mice receiving BM cells from PNC-27-treated donors (n=6) compared with those receiving BM cells from vehicle-treated donors (n=8). We also observed reduced hCD45⁺ (Figure S4I) and hCD45⁺CD34⁺ cell count (Figure S4J) in SP, and decreased SP weight (Figure S4K) in recipients receiving BM cells from PNC-27-treated donors compared to controls.

PNC-27 binding to membrane HDM2 induces poration and necrobiosis in AML cells

It has been reported that PNC-27 anticancer activity in solid tumor cells stemmed from the mechanism of membrane pore formation (so-called membrane poration) that causes cell death via necrobiosis rather than apoptosis.^12–18 Using electron microscopy, we showed membrane poration in PNC-27-treated AML cells, but not in PNC-29- or vehicle-treated AML cells at 24hrs (Figure S5A). This effect was followed by cell necrobiosis as supported by increased LDH levels in the culture media of PNC-27-treated mHDM2-positive MV4-11 cells but not in that of mHDM2-negative KG1A cells (Figure S5B). Increased cell death (i.e., Annexin V⁻ DAPI⁺ fraction) was observed following PNC-27 exposure in both mHDM2-positive MV4-11 and primary AML CD34⁺ cells, and not in mHDM2-negative KG1A and normal CD34⁺ cells (Figure S5C–D). Notably, no significant increase in apoptosis (i.e., Annexin V⁺ fraction) (Figure S5C–D) or in levels of pro-apoptotic proteins (i. e., p53, caspase-9, c-myc, bcl-2, bax, PARP-1, caspase-3) were observed (Figure S5E–F) in PNC-27-treated cells, thereby supporting cell death preferentially by necrobiosis rather than apoptosis.

PNC-27 enhances the interaction of HDM2 and E-cadherin on cell membrane

The molecular basis through which PNC-27-induced membrane poration remained to be fully elucidated. It has been reported that once bound to mHDM2, PNC-27 may induce poration by assuming a HLH structure formation in the membrane. However, if this was the only mechanism of action, one would expect necrobiosis to occur shortly after drug exposure. Instead, we observed cell death also occurring after 24hrs. E-cadherin (CDH1) is a structural membrane protein that plays a relevant role in cell migration and adhesion. This protein has been reported to colocalize with HDM2 on cell membranes in cancer cells and be negatively regulated by HDM2-mediated ubiquitination and degradation.^27,28 Interestingly, PNC-27 enhanced co-localization and interaction of E-cadherin and mHDM2 on the membrane of AML blasts, and not on that of normal CD34⁺ cells (Figure 4A–B, Figure S6A–B). This led us to hypothesize that PNC-27 may induce membrane poration via mHDM2-mediated E-cadherin ubiquitination and degradation.

Figure 4. PNC-27-mediated mHDM2 and E-cadherin interaction.

A. Co-localization of mHDM2 with E-cadherin (white signal) and mHDM2 with PNC-27 (yellow signal) in AML cells. Normal or AML cells were treated with 5-TAMRA fluorophore-labeled PNC-27 (25μM, 4hrs) or vehicle and stained with anti-HDM2 and anti-E-cadherin anti-bodies. Representative figures are shown. Scale bar, 5μm. B. Interaction of PNC-27 with mHDM2 and E-cadherin. Membrane fractions of PNC-27-treated AML cells were immunoprecipitated with control anti-IgG or antiPNC-27 (DO-1, to pull down PNC-27) antibody, then the immunoprecipitated complexes were immunoblotted with anti-HDM2 or anti-E-cadherin antibody. C. Effects of PNC-27 on the interaction between mHDM2 and E-cadherin. Membrane fraction of vehicle- or PNC-27-treated AML cells was extracted. The lysates were immunoprecipitated with anti-PNC-27 (top blot) or anti-HDM2 (middle and lower blots) and immunoblotted with anti-HDM2 (top and lower blot) or anti-E-cadherin (middle blot) antibody. D. Effects of PNC-27 on E-cadherin ubiquitination. AML cells were transfected with HA-Ubiquitin (Ub) followed by treating with or without PNC-27 (25μM, 18hrs). Lysate was immunoprecipitated with anti-E-cadherin antibody and immunoblotted with anti-HA antibody, densitometry quantification (convert to folds, related to loading control) of HA-Ub were indicated. E. Effects of PNC-27 on HDM2-regulated E-cadherin ubiquitination. Recombinant E-cadherin and HDM2 proteins were incubated together with E1, E2, ubiquitin, and ATP, in the presence or absence of PNC-27. Then each mixture was immunoprecipitated with anti-E-cadherin and immunoblotted with anti-ubiquitin. F. Effects of PNC-27 on mHDM2-induced E-cadherin degradation. Human normal and AML cells were treated with or without PNC-27 (25μM, 24hrs). Then total cell lysate was immunoblotted with anti-E-cadherin antibody (top) and lysate from membrane extract was immunoblotted with anti-HDM2 antibody (bottom). Full length and cleaved E-cadherin were indicated by arrow head. Densitometry quantification (convert to folds, related to loading control) of full-length E-cadherin were indicated. G-J. Effects of depletion of E-cadherin on cell necrosis. MV4-11 cells were transduced with shRNA control (shCtrl) or shRNA CDH1-1 or shRNA CDH1-2. Knock down of E-cadherin was confirmed by IB assays using E-cadherin antibody (G). Membrane poration, as indicated by arrow head, observed by transmission electron microscopy (H, scale bar, 0.5μm), and cell necrobiosis analyzed by LDH assays (I) and by flow cytometry using DAPI and Annexin V staining (J). Results shown represent mean ± s.e.m. ***P<0.001; by two-tailed, paired student’s t-test.

Thus, we analyzed membrane fractions from PNC-27- or vehicle-treated primary AML cells and normal cells using IB assays. Exposure to PNC-27 enhanced interaction of mHDM2 and E-cadherin compared to exposure to vehicle (Figure 4C) and increased levels of E-cadherin ubiquitination in AML blasts (Figure 4D), but not in normal cells (Figure S6C–D). To test whether PNC-27 enhanced E-cadherin ubiquitination via HDM2 E3 ubiquitin-protein ligase, we used a cell-free ubiquitination assay. We observed PNC-27 enhanced E-cadherin ubiquitination in the presence of HDM2 proteins (E3) with E1, E2, ubiquitin and ATP (Figure 4E). PNC-27 mediated enhanced degradation of E-cadherin was also observed in primary AML cells but not in normal cells (Figure 4F).

To further demonstrate the role of E-cadherin degradation in PNC-27 poration and in turn in antileukemia activity, we then transduced MV4-11 AML cells with lentivirus containing E-cadherin shRNA (i.e., shCDH1-1, or shCDH1–2) or control shRNA (shCtrl). We confirmed E-cadherin knock-down in MV4-11 cells transduced with shCDH1-1 and shCDH1-2 compared to those transduced with shCtrl (Figure 4G). We observed formation of membrane pores (Figure 6H), increased LDH levels (Figure 4I) and cell death in MV4-11 cells transduced with shCDH1-1 and shCDH1-2 (Figure 4J) but not in those transduced with shCtrl. In addition, we transfected THP-1 AML cells with siCDH1 and observed increased necrobiosis compared to siCtrl transfected control (Figure S6E–G). Since E-cadherin functional antibody (EA) binds to E-cadherin, we expected that EA would decrease the binding of E-cadherin and mHDM2 and in turn E-cadherin ubiquitination and degradation otherwise enhanced by PNC-27. Similarly, we expected a decrease in mHDM2-mediated E-cadherin degradation upon pre-treatment with a proteasome inhibitor (bortezomib, B). In fact, after exposing primary AML blasts to vehicle, PNC-27, EA, B alone and in combination, we showed that both EA and B decreased E-cadherin degradation (Figure S6H), and partly reversed PNC-27-mediated cell death in primary AML blasts (Figure S5I).

Discussion

The regulatory activity of the E3-ligase HDM2 on the tumor suppressor protein p53 is well recognized as a tumor suppression mechanism that activates DNA damage-induced apoptosis. Under physiologic conditions, HDM2 is expressed in the cytoplasm where it negatively regulates the activity of p53. Recent reports, however, indicated an unanticipated higher expression of HDM2 on the membrane of various solid tumor cells, compared to their normal healthy counterparts.^12–16 Herein we show that mHDM2 is specifically expressed on the cell surface of AML cells but not on that of normal HSCs or ALL blasts, with higher expression in LSC-enriched CD34⁺CD38⁻ subpopulation of AML. We also demonstrate an association of mHDM2 with quiescence, leukemia initiating capacity, and chemoresistance in primary AML CD34⁺ blasts. Although the mechanisms through which HDM2 is uniquely localized on the membranes of AML blasts remains unknown, it offers a unique opportunity for specific targeted therapy against AML blasts. Of note, while some heterogeneity in mHDM2 expression levels could be observed among distinct AML samples (Figure 1B), we found no association of mHDM2 expression and AML genotypes, but the sample size of the patients tested was relatively small (Table S1). Analyzing publicly available gene expression databases, we found a higher HDM2 expression in intermediate and poor risk AML cytogenetic groups compared to the favorable risk group (Figure S7). However, only HDM2 RNA expression data were available and whether HDM2 mRNA levels correlate with mHDM2 levels remains to be elucidated.

PNC-27 is a novel peptide comprised of a p53-mimicking 32-amino-acid sequence with the ability to engage with mHDM2 expressed on the surfaces of cancer cells. By binding to mHDM2, PNC-27 induces cell death via necrobiosis.^12–16 Herein we show that PNC-27 selectively kills AML blasts that express mHDM2 in vitro and in vivo but spares normal HSCs without this expression. Notably, we demonstrated that mHDM2 is also expressed in human CD34⁺CD38⁻ AML cells, which are reportedly LSC-enriched,²⁹ and in Mll^PTD/WTFlt3^ITD/ITD mouse AML LSK cells, which also represent a LSC-enriched cell subpopulation (unpublished data). AML cells with higher mHDM2 expression have a higher leukemogenic capacity and are more quiescent and chemoresistant than AML cells with lower mHDM2 expression. In secondary transplant experiments with mice engrafted with human AML blasts or Mll^PTD/WTFlt3^ITD/ITDAML LSK cells, we observed that survival was significantly prolonged in mouse recipients of BM from primary PNC-27-treated mouse donors compared with recipients of BM from primary PNC-29- or vehicle-treated mouse donors suggesting a PNC-27 antileukemia activity on leukemia-initiating cells (i.e., LSCs). More details PK/PD modeling studies to find the optimal dose/schedule of the drug are currently ongoing. As recent studies have also reported that PNC-27 enhances the activity of other antitumor compounds in solid tumors,³⁰ it would be interesting to test combinations of this peptide with other emerging and effective antileukemia agents (e.g., venetoclax) with different mechanisms of action. Of note, PNC-27 was selectively active on AML blasts as we showed that it did not affect the normal hematopoietic reconstituting capability of HSCs.

We also provide new insights into the mechanism of PNC-27 antileukemia activity. We showed pore formation on membrane of mHDM2-expressing AML cells after exposure to PNC-27 followed by necrobiosis-induced cell death. While it has been reported that PNC-27 induced necrobiosis is mediated by the HLH structure formed once PNC-27 binds to mHDM2^12–21 on the membrane, we were surprised that cell death continued to occur after 24hrs of drug exposure, thereby implicating other possible mechanisms. E-cadherin, a protein expressed on the membrane of normal and cancer cells, has been reported to function as a cell adhesion and migration factor.²⁸ In AML, reduced expression of E-cadherin has been shown to inhibit adhesion of leukemia cells to mesenchymal stem cells and leukemia growth.^31–33 While the role of E-cadherin in myeloid leukemogenesis remains to be further explored, it is known that HDM2 is reportedly a negative regulator of E-cadherin. ²⁷ Here we showed that PNC-27 enhanced protein-protein interaction between mHDM2 and membrane E-cadherin, which is then ubiquitinated and degraded, this coincided with membrane poration and ultimately necrobiosis of AML blasts.

In conclusion, we report here that mHDM2 is a novel, selective, and actionable target that is exclusively expressed on the surface of AML CD34⁺ blasts and LSC-enriched CD34⁺CD38⁻ population but not on the surface of normal CD34⁺CD38⁻ HSC. Furthermore, we showed that mHDM2 is associated with quiescence, leukemia initiating capacity, and chemoresistance in AML blasts. The absence of mHDM2 in normal HSCs offers a unique window of therapeutic opportunity for compounds like PNC-27 to eliminate AML blasts, including LSCs, while sparing normal HSCs.